|Publication number||US7640398 B2|
|Application number||US 11/178,713|
|Publication date||Dec 29, 2009|
|Filing date||Jul 11, 2005|
|Priority date||Jul 11, 2005|
|Also published as||US20070011405, WO2007008324A2, WO2007008324A3|
|Publication number||11178713, 178713, US 7640398 B2, US 7640398B2, US-B2-7640398, US7640398 B2, US7640398B2|
|Inventors||Vijay P. Adusumilli|
|Original Assignee||Atmel Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (13), Classifications (16), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates generally to nonvolatile memory devices, and more particularly to flash or EEPROM devices having two levels of internal cache.
Typically, a memory device will be coupled to an external control device such as a microprocessor. The microprocessor may be incorporated into a personal computer, a personal digital assistant, a telephone, a camera, or other device requiring a nonvolatile memory. A multitude of devices including PDAs, wireless devices, and cell phones continue to evolve and incorporate new multifunction capabilities. New capabilities include Web access, a digital camera, video, and music storage. To be marketable, these new devices must provide new capabilities at lower costs and in smaller spaces. In addition, nonvolatile memory devices must have higher capacities, improved speed, and improved interface flexibility.
For example, in the cell phone market, previous voice only cell phones utilized approximately 4 to 8 megabytes of memory to store data such as phone numbers, call logs, or messages. Currently, consumers now demand cell phones that are feature-rich. New cell phone devices now include Internet browsing, text messaging, games, Java applications, music, and digital cameras. These exemplary applications have caused an increase in memory requirements. Typically, cell phone manufacturers now use 64 to 256 megabytes or more memory to store large amounts of data including pictures and music.
Memory options when designing cell phones are numerous; a conventional memory architecture for a multifunction cell phone may use NOR flash for code storage, PSRAM for workspace, and NAND flash for data storage. Some designers also include SRAM for backup. NAND flash memory currently has the lowest cost per bit, however, NAND flash memory also has a slower random access time compared to other memory types and no capability for byte level programming.
A read access cycle time for NAND flash memory may be approximately 25 milliseconds. However, in typical applications, stored data is read into a page register and the data may be serially clocked from the memory device within a 50 nanosecond clock cycle. For example, U.S. Pat. No. 5,488,711 to Hewitt et al. describes a write cache for reducing the time required to load data into an EEPROM device. Although the architecture described by Hewitt et al. improves the performance of the memory device, further performance increases using different or improved architectures are possible.
A nonvolatile memory device utilizes two portions, or levels of cache to reduce the time it takes to read and write data. In particular, the cache and page register are configured so that read pages of data are copied to a first level of cache. Pages of data are read to fill the first portion of cache. When the first portion of cache is full, another page of data is read, and the data stored in the page register and the first portion of cache are copied to a second portion of cache. A read or write operation may then be performed at the same time that the pages of data in the second portion of cache are being copied to an input-output circuit and serially transferred to a device that is external to the memory device. (A serial transfer can also refer to bit/byte/word serial transfers.)
The control logic circuit 50 coordinates or controls the data transfer in the memory device. The control logic circuit 50 may be implemented as a state machine or a microcontroller or any sequential controller. In one embodiment, the control logic circuit 50 receives commands from a device that is external to the memory device 100. For example, a read command or a write command may be presented to the memory device 100 followed by an address or address range in the memory array 10. In response, the control logic circuit 50 controls word lines and bit lines coupled to the memory array 10 to address and read data from the memory array 10.
Additionally, the control logic circuit 50 coordinates or controls the transfer of data between the memory array 10 and the data register 20. The control logic circuit 50 also coordinates the transfer or copying of data between the data register 20 and L1 cache 31, the transfer or copying of data between the data register 20 or L1 cache 31 and L2 cache 32, and the control logic circuit 50 coordinates the transfer or copying of data between the L2 cache 32 and the input-output circuit 40. In one embodiment, the input-output circuit 40 contains a pipeline register.
During a read operation, data stored in the memory array 10 are transferred to the data register 20 via the sense amplifiers 11. The data register 20 is selectively coupled to the first level of L1 cache 31 and data temporarily stored in the data register 20 is copied to a selected portion of the L1 cache 31. Data continue to be read from the memory array 10 into the data register 20 and copied into the L1 cache 31 until the L1 cache 31 has been filled with data. Data stored in the L1 cache 31 are then copied to the L2 cache 32. Portions of the data stored in the L2 cache 32 are transferred or copied to the input-output circuit 40. The input-output circuit 40 then serially outputs the data, while the next read cycle is simultaneously being performed by the data register 20 and the L1 cache 31. (A serial transfer can also refer to bit/byte/word serial transfers.)
Alternatively, the input-output circuit 40 may also be directly coupled to the L1 cache 31 and data may be serially transferred directly from the L1 cache 31.
During a read operation, a page of data is read from the memory array 10 into the data register 20 and the data register page is copied to one of a plurality of L1 bit arrays 101, 102, 103 using a plurality of select devices 104, 105, 106. In one embodiment, a first page of data is read into the page register 20 and the data values are presented on a data bus 110. At least one of the select devices 104, 105, 106 couples the data bus 110 to a selected L1 bit array 101, 102, 103. For example, the select device 106 is activated coupling the data bus 110 to the L1 bit array2 103. The data register page is then copied to the L1 bit array2 103. At the same time, the select device2 105 and the select device1 104 do not actively couple the data bus 110 to the L1 bit array1 102 or to the L1 bit array0 101.
After the first data register page has been copied from the data register 20 to the L1 bit array2 103, the data register page is overwritten by a second page of data from the memory array 10. Pages of data continue to be read from the memory array 10 into the data register 20 until all of the L1 bit arrays 101, 102, 103 have had data copied to them and the L1 cache 31 is full of data. The second and third data register pages are copied from the data register 20 into the L1 bit array1 102 and the L1 bit array0 101. When the L1 bit arrays 101, 102, 103 are full of data, another read operation is performed and a page of data is read from the memory array 10 into the data register 20. In another embodiment, a data register page may be copied to any single selected L1 bit array 101, 102, 103 or copied to a plurality of L1 bit arrays. In an alternative embodiment, the first data register page is copied from the data register 20 directly to a single selected L2 bit array 201, 202, 203, 204 or copied from the data register 20 to a plurality of bit arrays in the L2 cache 32.
The data in the data register 20 and in the L1 cache 31 are then copied into the corresponding L2 bit arrays 201, 202, 203, 204. The page of data in the data register 20 is copied to the L2 bit array0 201 via the select device0 108, and the L1 bit arrays 101, 102, 103 are copied to the corresponding L2 bit arrays 202, 203, 204 in a single cycle. The data in the L2 cache 32 are then copied to an input-output circuit 40. The input-output circuit 40 then serially outputs the stored data, for example on a pin or line of the memory device 100, bit-by-bit to an external device such as a microprocessor (not shown).
In one embodiment, an entire page (four data register pages) of data is output word-by-word. A plurality of lines may provide multiple bits of data in parallel where the data are output with each line of the word providing a serial stream of data bits to an external device (not shown). For example, a 16-bit word of data is presented on 16 lines and each bit of the 16-bit word provides a serial stream of data so that an exemplary page of data at 528 by 16 is output to the microprocessor. In another example, the data in the input-output circuit 40 are presented to an external device as a 64-bit word (64 bits in parallel), serially clocking each bit of the 64-bit word for 264 cycles to present the entire page of data to the external device or microprocessor. Alternatively, any number of bits in parallel may be presented to an external device. Additionally, in other embodiments, the data lines may include additional bits such as error-checking codes or error correction bits.
The multiplexer 310 selectively couples the L2 bit array0 201 and the L1 bit arrays 101, 102, 103 to the data register 20. After the input-output circuit 40 has provided enough data to fill the L2 bit arrays 201, 202, 203, 204, the entire page data in the three L2 bit arrays 202, 203, 204 are copied to the corresponding L1 cache bit arrays 101, 102, 103. The multiplexer 310 selectively couples the L2 bit array0 201 to the data register 20 via multiplexer select line(s) 311 and the page of data in the L2 bit array0 201 is copied to the data register 20. A first write operation is performed to the memory array 10 while the L2 bit arrays 202, 203, 204 are being copied to the L1 bit arrays 101, 102, 103.
In one embodiment, the L2 bit arrays 201, 202, 203, 204 are set to a predetermined value. New data are then copied into L2 bit arrays 201, 202, 203, 204 from the input-output circuit 40, and simultaneously, pages of data in the L1 cache 31 are copied to the data register 20 and written to the memory array 10 (
When the L2 cache 32 is full, a second determination 640 is made whether the L1 cache 31 has completed any previous operation and is available 640. If the L1 cache 31 is not available, data in the L2 cache 32 are held and in one embodiment, a predetermined wait period is executed 650 or alternatively, a “no op” instruction is performed until the L1 cache 31 is available. When the L1 cache 31 is available, data in the L2 cache 32 are copied 660 into the L1 cache 31. Next, data in the L1 cache 31 are copied 670, page by page, to a data register 20 and written page by page from the data register to a memory array 10, while simultaneously inputting 610 additional data, and copying 620 the additional data to the L2 cache 32, 620 until a determination 630 is made that the L2 cache 32 is full.
Those of skill in the art will recognize that the present invention can be practiced with modification and alteration within the spirit and scope of the appended claims and many other embodiments will be apparent to those of skill in the art upon reading and understanding the above description. For example, one skilled in the art will recognize that data transfers and copies can be bit-by-bit, word-by-word, or page-by-page. A skill artisan further would recognize that an array in the present invention is not limited to a particular page size. The number of L1 and L2 bit array pages may differ compared to the above embodiments and examples. In addition, other embodiments of the input-output circuit 40, the L1 cache 31, and the L2 cache 32, may be implemented using a variety of page sizes to transfer or copy pages of data. Also, the L1 and L2 cache pages may be a single cache memory, having multiple pages that may be flexibly controlled. In addition, the select devices coupled to the first level of cache (L1) for a read operation may also be incorporated or coupled to the circuit described to perform a write operation, and the read and write operations described can be performed by a single circuit arrangement. The description is thus to be regarded as illustrative instead of limiting. The scope of the invention should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which said claims are entitled.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4860192||Oct 3, 1986||Aug 22, 1989||Intergraph Corporation||Quadword boundary cache system|
|US5161219||May 31, 1991||Nov 3, 1992||International Business Machines Corporation||Computer system with input/output cache|
|US5488711||Apr 1, 1993||Jan 30, 1996||Microchip Technology Incorporated||Serial EEPROM device and associated method for reducing data load time using a page mode write cache|
|US5699317||Oct 6, 1994||Dec 16, 1997||Ramtron International Corporation||Enhanced DRAM with all reads from on-chip cache and all writers to memory array|
|US5850534||Jun 5, 1995||Dec 15, 1998||Advanced Micro Devices, Inc.||Method and apparatus for reducing cache snooping overhead in a multilevel cache system|
|US6209067 *||Dec 4, 1995||Mar 27, 2001||Compaq Computer Corporation||Computer system controller and method with processor write posting hold off on PCI master memory request|
|US6347063||Jul 17, 2000||Feb 12, 2002||Ubishi Denki Kabushiki Kaisha||Semiconductor memory device including dynamic type memory and static type memory formed on the common chip and an operating method thereof|
|US6604171||Sep 29, 2000||Aug 5, 2003||Emc Corporation||Managing a cache memory|
|US7073016 *||Oct 9, 2003||Jul 4, 2006||Micron Technology, Inc.||Random access interface in a serial memory device|
|US7366044 *||Jun 21, 2006||Apr 29, 2008||Kabushiki Kaisha Toshiba||Systems and methods for data transfers between memory cells|
|US20040193782||Mar 25, 2004||Sep 30, 2004||David Bordui||Nonvolatile intelligent flash cache memory|
|US20050223158 *||Apr 5, 2004||Oct 6, 2005||Sun-Teck See||Flash memory system with a high-speed flash controller|
|US20050268025 *||May 27, 2004||Dec 1, 2005||Peter Smith||Configurable ready/busy control|
|U.S. Classification||711/122, 711/129, 711/103, 711/165, 711/162|
|Cooperative Classification||G11C16/26, G06F12/0893, G06F2212/2022, G06F12/0897, G11C2207/2245, G11C7/1039|
|European Classification||G11C7/10M5, G11C16/26, G06F12/08B22, G06F12/08B22L|
|Sep 19, 2005||AS||Assignment|
Owner name: ATMEL CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:ADUSUMILLI, VIJAY P.;REEL/FRAME:016994/0498
Effective date: 20050708
|May 29, 2013||FPAY||Fee payment|
Year of fee payment: 4
|Jan 3, 2014||AS||Assignment|
Owner name: MORGAN STANLEY SENIOR FUNDING, INC. AS ADMINISTRAT
Effective date: 20131206
Free format text: PATENT SECURITY AGREEMENT;ASSIGNOR:ATMEL CORPORATION;REEL/FRAME:031912/0173